This final aim has been scarcely investigated by the Scientific Community by now.
This paper is organized as follows: Section 2 will be dedicated both to the analysis of the state-of-the-art of integrated metric
survey techniques used for 3D modeling reconstruction, taking into account both the advantages and the disadvantages. Section
3 describes the case study, the survey and the modeling technique, while the following Section 4 describes the
relationship between BIM and restoration with a specific regard to the modeling of the decays. Section 5 is dedicated to the
conclusions and final considerations.
2. STATE OF-THE-ART:
FROM BIM
TO HBIM.
ADVANTAGES AND
CRITICALITIES
For several years, most countries in Northern Europe have widely implemented BIM methodology, particularly in the domain of
new constructions. Several papers deal with the description of the methodologies, processing steps and standards on different case
studies in the field of BIM
Eastman et al., 2011. Large activities on BIM implementation have been developed in foreign
Countries such as United Kingdom BSC, 2013; AEC, 2012, Finland
Building smart Finland, 2012, and Norway
Statsbygg, 2013.
The development of this emerging approach is slower in the South of Europe: only in the last years, a large activity around
this topic is going to affirm the increasing interest on the BIM processes. In Italy Country deeply characterized by existing
historical buildings and historical contexts, BIM is becoming a very important topic. We are assisting to an increasing
development of HBIM
Historical Building Information Modeling, defined by the scientific literature as the most
interesting method to define a semantic aware 3D reconstruction of historical elements, regardless the needs, the aims and the uses
of a 3D model based on parametric approach. The modelling and monitoring phases have been analyzed by several authors:
Barazzetti et al., 2015a; Dore et al., 2015; Baik et al., 2014; Oreni, 2014; Ludwig et al., 2013, Miller et al., 2014, Volk et al.,
2014; Barazzetti et al., 2015b; Quattrini et al, 2015; Barazzetti et al., 2014; Lin, 2014, Apollonio et. al, 2012; Donato et., al 2017;
Castagnetti et al., 2017.
HBIM was developed in the last ten years Murphy et al, 2009
like a new prototype-system of BIM, a modelling of historic structures as parametric objects in a database “library”. These
objects can be collected and modelled using a reverse engineering process, creating full 2D and 3D models including
details behind the objects surface, concerning its methods of construction and material makeup. Information are taken on the
base of 3D reality-based data from point clouds, by automatic or semi-automatic recognition Murphy et al, 2012.
The chance to combine a high level of different information in a unique and shared database allows figuring out the next spread of
the HBIM adoption as a standard instrument for all the decision- making processes related to the Cultural Heritage domain. Beside
the geometric 3D information, it is possible to define all the alphanumeric information of an historical complex. The main
important
are: stratigraphy,
decay analysis,
structural information, material composition, technological features,
previous surveyinterventions, old and recent pictures, historical documents, etc. The different category of information are
hierarchically organized, and, creating a full-integrated survey, can be used and managed for several aims.
According to these assumptions, HBIM could be defined as a semantic-aware database connected to the historic buildings, in
which the geometric model need to be join to descriptive information multi-source Quattrini et al, 2015. Therefore, it is
necessary to start from a correct survey to correctly define the architectural shapes. Actually, the most used strategy is based on
the point clouds obtained from range base LiDAR, SLAM etc or image-based techniques SfM, photogrammetry. Moreover,
Geomatics techniques allow to easily generate 3D models, orthophotos and other outputs with assessed accuracy.
The subsequent conversion of 3D information into parametric components, especially in an historical background, is very time-
consuming and has several open issues Lo Turco, Sanna, 2010. According to the geometric rules defined by historical treatises
andor the normal practice of the past, the architectural shapes have to be recognized and segmented from the raw 3D data to be
embedded in the HBIM: surfaces planes, curves, or extrusion, volumes and complex objects. This conversion could be based
on:
- manual modelling operations of volumes and shape from- scratch;
- semi-automatic procedures, by managing cross sections and surface extrusions;
- automatic procedures, to rebuild the object according to automatic surface extraction from the point cloud.
The gap between the transformations of a point cloud into a parametric model is today a bottleneck into HBIM adoption
because of very heterogeneous geometries: most of them need to be manually re-modelled because no automatic recognition can
be used for the great metric discrepancy between real objects and virtual reconstructions. Within this scenario different strategies
and approaches were developed in the last years, thanks both to internal tools linked to the BIM software or external software that
are useful to simplify the BIM modelling phase.
Regarding the application of BIM to restoration interventions, the attention was focused on the chance to model degradations in the
BIM platform, enriching the related database with graphic, geometric and alphanumeric data that can be effectively used to
design and manage future interventions. In the next sections a complete description of the achieved results is widely explained,
with a final focus on the proposed approach followed for the graphic representation of pathologies. This issue is very complex
to be managed within the HBIM environment; it was proposed to use an adaptive component, a specific object that can be adapted
to a surface using movable vertex. This flexible geometry could be associated with flat and curved surfaces as well. This is
considered by the authors as one the most innovative part of the entire work, because no instances related to these themes were
found Lo Turco et al., 2017.
3. THE CASE
STUDIES: PALAZZO
SARMATORIS AND
TORINO SMISTAMENTO
ROUNDHOUSE
Two different tests are here presented: the first, Palazzo Sarmatoris, a historic complex dated back to the XVII century.
The second, Smistamento Roundhouse in Torino a historical industrial building erected at the beginning of the XX Century.
Below, some information related to the buildings, then the description of surveys methodologies and the strategy employed
for the HBIM realization are reported in order to obtain a final product that could be a useful instrument for documenting and
improve the knowledge of the analyzed artifacts.
3.1 Historical background
The area of Palazzo Sarmatoris was reported in the Theatrum Sabaudiae as Palatium Sarmatoris Figure 2, Roccia, 2000. The
building is located in Salmour, a little village close to Fossano in the Cuneo province. Despite the important historic interest, the
Palazzo Sarmatoris is abandoned; and because of that, several thefts of the historical furniture occurred in the last years.
This contribution has been peer-reviewed. doi:10.5194isprs-archives-XLII-2-W5-605-2017 | © Authors 2017. CC BY 4.0 License.
606
Nowadays, according to the lack of ordinary maintenance the structures, the external and internal walls present several problems
and decays: without an important restoration project and the connected retrofit works probably the structure will suffer
important damages and significant breaks in the next future.
Figure 2. A view of the Palazzo Sarmatoris in the
Theatrum Sabaudiae
Roccia, 2000. The second building is a property of the Ferrovie dello Stato, now
abandoned and originally used for the storage of locomotives; during the last years, a specific intervention has been proposed. The
building was built to adapt the railways for the Universal Exhibition of 1911, to celebrate the 50th anniversary of the
establishment of the Unification of Italy. The maintenance area was constituted by a roundhouse
containing 51 shelter railroad tracks next to two sheds for the most complex interventions A and B parts, Fig. 3. The
bombings of 1943 caused huge damages to the circular structure, knocking down some of the 51 aisles, reducing the overall
number to 32.
Figure 3. Detail of the site plan, developed in 1905.
3.2 3D metric survey
In order to realize a parametric model in a HBIM environment the objective of the survey need to be the realization of a
complete 3D point cloud of the analyzed objects. The surveys operations have been carried out by using the image base
approach photogrammetry for the Palazzo Sarmatoris and TLS techniques for the metric survey of the Smistamento
Roundhouse. In the first case a complete photogrammetric survey has been
achieved using terrestrial and UAV images oblique and nadir images, in the external part and in the main internal rooms the
other rooms were surveyed by manual measurements integrated by preexisting drawings. In the second case, by considering the
evident symmetry of the building and the accuracy requirement of the survey a complete external survey has been done and only
one module of the structure was also surveyed indoor by using TLS instruments.
In both cases, a first order control network has been realized by using GNSS andor Total Station survey
, while the GCPs needed for image orientation and TLS point clouds registration were
measured by using a Total Station connected to the existing first order control networks.
The Palazzo Sarmatoris façades and the roof of the ancient Palace were acquired by means of a COTS Commercial Off The Shelf
UAV system able to acquire nadir and oblique images as well. Moreover, in order to improve the level of detail for better
documenting the lower parts of the façades and the main internal room, the UAV data were integrated with close-range image
acquisition. The well-known Phantom 4 produced by DJI acquired the aerial
data. The flight plans were realized at an elevation of about 15 m over the ground level in order to have a GSD Ground Sampling
Distance of about 4 mm. Finally, a nadir 3 strips and oblique acquisition around the Palace with the camera oriented at 45° have
been realized for better documenting the area Figure 4.
Figure 4. Scheme of the UAV acquisition nadir stripes left, oblique acquisition route right of Palazzo Sarmatoris.
Figure 5. The Achieved point cloud obtained using UA Vand terrestrial data in the external part of Palazzo Sarmatoris.
Figure 6. Point clouds of the five main room in the lower floor. of Palazzo Sarmatoris .
This contribution has been peer-reviewed. doi:10.5194isprs-archives-XLII-2-W5-605-2017 | © Authors 2017. CC BY 4.0 License.
607
The terrestrial acquisitions were recorded using a Nikon 5D full frame digital camera equipped with a lens of focal length 17mm.
In the external part the shooting distance was about 4m with as a result a GSD of about 2 mm. The data were processed using
Agisoft Photoscan combining the different dataset using the common reference system measured with traditional topographic
survey operations. The mean RMS measured on several Check Points CP in the internal and external part was less than 1 cm.
In the following figures a view of the realized point clouds of the external part Figure 5 and in the main internal rooms Figure 6
of Palazzo Sarmatoris are presented. The Smistamento Roundhouse was surveyed using a TLS,
following the typical approach for architectural heritage Axelsson, 1999; Vosselman, Maas, 2010
. A Faro Focus 3D
was employed for acquiring the complete area. All the building was recorded in the external part with a quality of 15 that
according to the typical settings of the Faro scanner means one point each 8 mm at 10 m. Only two internal parts were surveyed
with the laser because the structural symmetry was judge enough to allow the accuracy requirement satisfaction.
After the acquisition, all the scans were colored, filtered, registered and referenced according to the first order control
network previously realized. Figure 7 shows a view of the complete point cloud with the first
order control network measured with GNSS and Total Station
measurements.
Figure 7. First order control network above and the registered point clouds below.
The result of the registration phase, performed by using the cloud to cloud approach Iterative Closest Points and furthermore
using the Ground Control Points for georeferencing the products has been checked on some CP. The mean obtained RMS was
under 2 cm.
3.3 From the point cloud to the parametric model